In situ measurements of the electrical conductivity of

In situ measurements of the electrical conductivity of bismuthmolybdate catalysts in operation for oxidativedehydrogenation of buteneCitation for published version (APA):van Oeffelen, D. A. G., Hooff, van, J. H. C., & Schuit, G. C. A. (1985). In situ measurements of the electricalconductivity of bismuth molybdate catalysts in operation for oxidative dehydrogenation of butene. Journal ofCatalysis, 95(1), 84-100. https://doi.org/10.1016/0021-9517%2885%2990010-7, https://doi.org/10.1016/0021-9517(85)90010-7

DOI:10.1016/0021-9517%2885%2990010-710.1016/0021-9517(85)90010-7

Document status and date:Published: 01/01/1985

Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne

Take down policyIf you believe that this document breaches copyright please contact us at:[email protected] details and we will investigate your claim.

Download date: 04. Apr. 2022

https://doi.org/10.1016/0021-9517%2885%2990010-7

https://doi.org/10.1016/0021-9517(85)90010-7

https://doi.org/10.1016/0021-9517(85)90010-7

https://doi.org/10.1016/0021-9517%2885%2990010-7

https://doi.org/10.1016/0021-9517(85)90010-7

https://research.tue.nl/en/publications/6f4451c6-455d-4ef3-af83-77c6403db838

JOURNAL OF CATALYSIS 95,84-KtO (1985)

In Situ Measurements of the Electrical Conductivity of Bismuth Molybdate Catalysts in Operation for Oxidative

Dehydrogenation of Butene

D. A. G. VAN OEFFELEN,* J. H. C. VAN HOOFF,* AND G. C. A. SCHUIT~

*Laboratory for Inorganic Chemistry and Catalysis, Eindhoven University of Technology, P.0. Box 513, 5600 MB Eindhoven, The Netherlands, and tCenter for Catalytic Science and Technology,

Department of Chemical Engineering, University of Delaware, Newark, Delaware 1971 I

Received February 21, 1984; revised March 23, 1985

In situ measurement of electrical conductivities were performed on MoOz, Bi2MoZ09. and B&Moo6 (koechlinite), the latter either as such or doped with small amounts of MO or Bi; the reaction applied was the oxidative dehydrogenation of butene to butadiene. The reaction was performed in continuous flow or by pulsing butene over the catalyst, often followed by the addition of Or pulses. The catalysts were characterized by surface area, XRD, and XPS. The latter measurements showed that surface Bi/Mo ratios were often different from those in the bulk; reduction by butene at temperatures around 673 K sometimes led to considerable changes in the surface Bi/Mo ratio that ran parallel with changes in activity. Pulsing butene in a He carrier stream over the catalyst strongly increased the electrical conductivity while pulsing O2 decreased it again; the conductivity is almost entirely located in the surface layers. The koechlinite-type catalysts, when heated in He, showed a considerably higher conductivity than the other catalysts. This is ascribed to a dissociation of O2 presumably from the B&O2 layers. Subsequent pulsing of butene first increased the conductivity but for catalysts with Bi/Mo bulk ratios > 1, the conductivities became constant while reduction continued. The first process is supposed to be connected with a stripping of the oxygen in the surface layer, the second with a migration of O*- from bulk to surface and of electrons from surface to bulk. A model was presented for the surface band. Q 1985 Academic PESS.

Inc.

INTRODUCTION

The essence of the redox mechanism as accepted for the catalytic oxidation of olefins over oxide catalysts, is a donation of protons and electrons by the hydrocarbon to the solid catalyst. The protons combine with oxygen atoms to form water thereby depleting the surface anion concentration; 02 molecules replenish surface anions by dissociative adsorption. The state of the electrons in the solid is not very well known and one might hope to obtain some information on this state by studying the electron mobility in an electric field, i.e., mea- suring the electrical conductivity of the catalyst. A well-known difficulty in analyzing the experimental conductivity data is given by the contribution to the overall conductivity of the migration of ions.

Several reports have appeared in literature covering studies about the electrical conductivity of bismuth molybdates.

Pluta (I, 2) investigated the conductivity of Mo03, B&03, and bismuth molybdate (Bi/Mo = 0.7 on Si02 carrier) in air, oxygen, and propene and at temperatures between 400 and 500°C. Already during out- gassing of the bismuth molybdate, changes were observed, which were ascribed to de- sorption of chemisorbed oxygen atoms. Contact with propene raised the conductivity, while air (or oxygen) restored it to its original value. The increase in conductivity during propene admission was ascribed to a reduction of the oxide, while the decrease after oxygen addition was considered to be caused by the reoxidation.

Peacock et al. (3) also investigated the conductivity of pressed disks of Bi203,

84 0021-9517185 $3.00 Copyright 0 1985 by Academic Press, Inc. All rights of reproduction in any form reserved.

ELECTRICAL CONDUCTIVITY OF BiMoO CATALYSTS 85

Mo03, and bismuth molybdate (Bi/Mo = 0.7) in propene and air (or oxygen) at 500°C. By admitting propene to bismuth molybdate they observed a considerable increase in conductivity; this could be elimi- nated by subsequent admission of oxygen. They showed that the increase in conductivity was brought about by extensive reduction of the surface layers of the catalysts. Similarly, the decreases after oxygen admission had to be ascribed to reoxidation of the reduced surfaces. They were able to show that the bismuth molybdate catalyst was about 10% reduced when operating with mixtures of propene and oxygen under normal steady-state conditions. Further- more they were able to determine the temperature dependence of the conductivity of the catalyst in oxidized and in partially reduced states. They showed that the energy gap decreased with increasing reduction. They concluded that the conductivity measurements can produce information about the reduction of the surface and perhaps a few subsurface layers, but do not furnish information about the oxidation states of the individual ions present in the surface.

Morrison (4) described the interaction of bismuth molybdates (Bi/Mo = 0.7) with oxygen and propene and the corresponding conductivity changes. At T < 275°C propene reacted only with what is assumed to be adsorbed oxygen, so that the conductivity after propene admission reaches a maximum. However, at T > 275°C propene reacted also with “irreversible oxygen,” i.e., lattice oxygen, so that in this case the conductivity did not reach a maximum and reduction was continuous.

Our conductivity measurements on the bismuth molybdates differ from those men- tioned before on the following points.

The experimental arrangements were as follows. We first determined activities of the various samples as a function of the temperature in continuous-flow modes when the reactants streamed through a bed of catalyst powder.

We then repeated the experiments in a

reactor where a pellet was present so that the reactants streamed around the pellet; its electrical conductivity could then be measured either in AC or DC current. These experiments were not particularly informa- tive; electrical conductivities were low and in the order of 10-6-10-7 (a-* cm-l) when no butene was present to increase somewhat when the olefin was added to the gas. It was then decided to use the method whereby pulses of butene and oxygen were injected into a He stream passing the pellet. The electrical conductivities were measured when the olefin pulse passed the pellet while its degree of conversion was then measured chromatographically. To obtain an impression of how the differences in flow patter, i.e., gas stream flowing along the pellet or through a bed of powdered catalyst, influenced the conversions, separate experiments with pulses flowing through a bed were also performed; for obvious rea- sons the latter experiment did not allow measurements of the conductivity.

EXPERIMENTAL

Catalytic Activity Measurements

When the butene oxidation was studied in a continuous-flow apparatus, a constant flow of a mixture of 1-butene and artificial air (ratio C4Hs: 02:He = 20:20: 80), passed through the catalyst bed (400 mg, particle size 0.2-0.4 mm) that was placed in a microreactor made of quartz glass. The reactor was heated by means of a heating wire, mounted on an aluminum tube. The temperature inside the catalyst bed was measured by a chrome1 alumel thermocouple [see Batist et al. (6)].

Samples of the product stream were ana- lyzed gas chromatographically.

We used the following definitions:

activity = (buten& - (butenesh . 1w (buteneslo 0,

(diene) selectivity = (diene) + t(co,) * lOO%,

86 VAN OEFFELEN, VAN HOOFF, AND SCHUIT

FIG. 1. Experimental set up for the electrical conductivity measurements. S = sample, F = constant voltage supply, A = multimeter, V = spot galvanome- ter, J = junction box.

in which

(butenes& = concentration of butenes before reaction,

(butenes)r = concentration of butenes af-

(diene) W2)

ter reaction at temperature T, = concentration of butadiene, = concentration of carbon diox-

ide.

For pulse experiments we used the same apparatus but the reactor was placed between sample valve and gas chromatographic column. The helium carrier gas was passing continuously over the catalyst; by means of a sample valve it was possible to pulse amounts of 1-butene over the catalyst with subsequent analysis of the reaction products. The volume of the pulse was 0.6 cm3 (25°C 1 atm).

Electrical Conductivity Measurements The electrical conductivity was mea-

sured on samples which were pressed to rectangular pellets under a pressure of 1000 kg/cm2. The dimensions were about 15 x 5

x 5 mm. The weight of the pellet was about 0.8 g.

A four-probe technique was used, in which two of the electrodes served as current leads and the other two as voltage leads. Figure 1 gives a scheme of the experimental set up.

From current and voltage measurements the resistivity could be calculated. From the sample dimensions we could get a specific resistivity and an electrical conductivity:

1 L 1 a=i=x-R’

in which

L = distance between electrodes 3 and 4 (cm),

A = cross-sectional area of pellet (cm2), R = resistivity (n), p = specific resistivity (Cn * cm), (T = electrical conductivity (a-l * cm-‘).

The polarity could be changed to check charging effects.

The sample was spring-loaded between Pt contacts. Pt was used as electrode material; Pt leads were shielded by ceramic tubes. The electrode assembly was placed in a quartz glass tube, which could be sup- plied with different gases (Fig. 2). The outlet was connected with a gas chromatographic column, the same as used for the activity measurements. By means of a sample valve it was possible to carry out pulse experiments by pulsing 1-butene in a helium flow. The reactant gases were entering the tube directly below the catalyst pellet. The volume of the pulse was 0.9 cm3.

Because of the rather large volume of the tube it was not possible to get an ideal plug flow of the gases, thus preventing a good gas chromatographic analysis of the products. Freezing the reaction products for 15 min in a liquid nitrogen trap and subsequent quick heating to room temperature enabled us to get a reproducible analysis of the reaction products.


FIG. 2. Electrical conductivity cell and sample holder. a = Sample, b = electrodes, c = alumina ceramic tubes, d = stainless-steel holder, e = quartz glass tube, f = springs, g = inlet gases, h = exit gases, i = electrode connections, j = thermocouple.

XPS, XRD, and SA Measurements

X-Ray photo electron spectroscopy. The apparatus used was at the Physical Chemis- try Laboratory of the University of Gro- ningen (The Netherlands). It was an AEI ES 200 spectrometer with MgKa radiation (1253.6 eV), with the power supply run at 16 kV and 15 mA. The spectra were col- lected using a PDP 8 computer and stored in a 32K disk. The powdered samples were either stuck onto a piece of tape or pressed into a copper gauze. Sensitivity factors used were as given by Berthou and Jorgen- sen (5).

X-Ray difSraction. We used a Philips PW 1120 spectrometer with CuKa radiation and Ni filter.

Surface area measurements. These were performed with an areameter of Striihlein, which is based on the single-point BET method using nitrogen as adsorbate.

Preparation of the Catalysts

The preparations were carried out with the following chemicals: basic bismuth ni-

trate BiONO3 (Merck), powdered molybdic acid (BHD), ammonium heptamolybdate (NH&lvl0~0~~ * Hz0 (Merck), bismuth nitrate Bi(NO& * 5HrO (Merck).

BiaOs was prepared by heating basic bismuth nitrate, containing 79.94 wt% B&O3 at 520°C for 10 h to give ivBizOr (Sample A).

MOO, was prepared by heating molybdic acid, containing 88.98 wt% MoOj at 500°C for 4 h (Sample B).

Bi2MoOs was prepared according to the slurry method of Batist (6): 16.177 g molyb- die acid (0.1 mol MOOS) and 58.289 g basic bismuth nitrate (0.1 mol B&O,) were added to 1 liter boiling water. To the slurry was added a 4-cm3 volume of concentrated ni- tric acid, after which the reaction was started. The slurry was vigorously stirred. After 12 h of reaction the yellow colored mass was filtered off, dried at 120°C for 16 h, and calcined at 500°C for 2 h. The reaction could be carried out quantitatively as was shown by examining the obtained weight (Sample C).

Bi2Mol,cnOa.os: The same method was used as for BizMo06, except that we used


16.500 g molybdic acid (0.102 mol Mo03). Here again the reaction was quantitative (Sample D).

%&fo106.~: Again a slurry method was used, starting with 16.177 g molybdic acid (0.1 mol Mo03) and 59.204 g basic bismuth nitrate (0.102 mol Bi20X). The molyb- die acid was first dissolved in aqueous ammonia. After 18 h of stirring and boiling the mass was filtered off, dried at 120°C for 16 h, and calcined at 500°C for 2 h (Sample E). This preparation also yielded stoichiometric amounts as could be shown by weight analysis.

Bi2M0209: To 24.253 g Bi(NO& * 5H20 (0.025 mol Bi203) was added 13 cm3 concentrated HNOs and dissolved in 100 ml water; 8.827 g (NH4)6M07024 . 4H20 (0.05 mol Mo03) dissolved in 100 ml water was added to the first solution with stirring. By adding aqueous ammonia the final pH was adjusted to 4. Meanwhile the solutions were cooled in ice baths. The white precipitate was filtered off, dried at 120°C for 16 h, and calcined at 500°C for 2 h. This procedure yielded a stoichiometric sample as checked by weight analysis (Sample F).

Bi2MoX0i2: 14.577 g basic bismuth nitrate (0.015 mol Bi203) and molybdic acid (0.075 mol Mo03) were brought into 1 liter water. The slurry was heated and stirred for 30 h, during which time the color remained white. After filtration, drying at 120°C for 16 h, and calcining for 2 h at 480°C the resultant material contained the stoichiometric amounts of bismuth and molybdenum as was checked by weight analysis (Sample G) .

RESULTS

Activities and Selectivities

All the catalysts except A and B were tested for their activity and selectivity in the oxidative dehydrogenation of 1-butene under continuous flow. Conditions used were 400 mg catalyst, total gas flow 120 cm3/min, C4Hs:02: He = 20:20:80 (the results are given in Table 1).

The sample Bi2.&‘f006.M was totally inactive over the sample temperature range investigated (380-440°C).

The sample Bi2M01.0206.M showed a very high activity at temperatures higher than 420°C whereas the sample Bi2MoOd had already a high activity at 38O”C, but this latter catalyst showed a considerably lower selectivity.

The catalyst Bi2M0209 showed fairly high activity and a very high selectivity. The sample Bi2M03012 prepared had low activity and high selectivity.

At 440°C the order of activity per unit of weight was: Bi2M01,0206.~ > B&M006 > Bi2M0209 & Bi2M030i2 > Bidi!f006.o+

The order of selectivity at 440°C was: Bi2 Mo209 = Bi2M03012 = Bi2M01.&&.~ > Bi2 MO06.

The inactive Bi2.~M0G6.06 became reacti- vated after adding Mo03. This was accom- plished by taking 375 mg Bi2.04M0G6.~ and adding 25 mg Mo03, corresponding to about 25 mol% MO extra. After 30 min at T = 440°C in the reactor, this mixture exhib- ited an activity of 59% and a selectivity of 89.5%. At the end of this reaction time we could observe an appreciable amount of Bi2 M03O12 to be present, as shown by the in- frared spectrum [bands at 950,930, and 900 cm-l, which are attributed to the 2/3 compound (6)].

It was also possible to activate the Bi2.m MoO~.~ sample by addition of an amount of a Bi2M030i2 sample with low activity.

The mixing of 308 mg Bi2.~M0G6.~ and 100 mg Bi2M030i2 after 30 min in the reactor gave a catalyst with an activity of 47% and a selectivity of 89% at 440°C.

XPS Measurements We used the values of Berthou and

Jorgensen (5) for the elemental sensitivities of Bi(4f) = 4.3, Mo(3d) = 1.6, O(ls) = 0.6, and as reference C(ls) = 2.85 eV. Table 2 lists the binding energies (standardized) and intensities for the catalysts given before. Table 3 compares the bismuth-molybdenum ratios for bulk compound with the ob-


TABLE 1

Comparison of Conversions for Various Bi Molybdate Catalysts in Diierent Flow Situations

Catalysta Catalyst composition

(Bi/Mo ratio)

Bulk Surface

Reaction temp. W)

Butene conversion (%)

Continuous Pulse method I Pulse method II flow

Pulse I pulse (n)b Pulse 1 Pulse (n)b

E, 10.7 350 2.04 2.27 400

450

c, 7.9 350 2.00 2.11 400

450

D, 7.3 350 l.% 1.78 400

450

F, 4.3 350 1.00 0.93 400

450

G, 2.3 350 0.67 0.70 400

450

a Symbol and surface area (in m*/g). b Pulse (n) = highest conversion at pulse n. c Selectivity = 84%, all others 295%.

0 0 0

79’ 79 83

<26 54 95

<9 31 72

<18 ~18

18

0 0 3

25 44 45

18 24 40

12 20 30

0 14(8) 4W

All lower All lower

W)

58(9)

WI 51(5)

1 3

32

3 40 48

23 33 42

17 26 36

4 10 17

17(5) 39(5)

13(5) All lower All lower

All lower All lower All lower

served XPS values for its surface. Figure 3 some powdered samples on tape. A shoul- shows bismuth and molybdenum signals for der can be observed at higher binding ener- Bi2Mo06 and Fig. 4 gives oxygen signals for gies; the low BE belongs to lattice oxygen.

TABLE 2

XES Data: Binding Energy Values (eV) and Intensities (Int.) of Bi, MO, and 0 Signals for the Various Catalysts

Sample Bi 4f MO 3d 0 1s

b.e. (eV) bit. b.e. (eV) bit. b.e. (eV) bit.

Bi2.&G..M

BiN+ .m06.w

Bi,MoQ

Bi,MqO,

BiZMojOu

Blz.orMoO,.,

(after activation)

159.3 164.6 159.5 164.9 159.8 165.1 160.0 165.5 159.9 165.4 159.3 164.6

8.38

10.65

131.7

28.46

3.169

166.76

232.6 235.7 232.7 235.9 233.0 236.2 233.0 236.0 233.2 236.1 232.6 235.7

3.73

6.03

62.4

30.68

4.557

151.3

530.0 17.41 531.7 (sh) 530.3 25.75 532.1 (sh) 530.6 295

530.8 126.7

530.7 14.%

530 403.4


TABLE 3

Comparison of the Calculated Bismuth-Molybdenum Bulk Ratios with the Observed Values

from XPS Experiments

Sample Calc. XPS

B1zMo1.ozQm Bi2.dfo06.M B&Moos Bi2M0209 Bi2Moj0t2 BlmMO%x

(after activation)

1.96 1.767 2.04 2.245 2.00 2.11 1.00 0.928 0.67 0.695

1.10

Pressing in Cu gauze showed the high BE signal only as broadening.

Pulse Experiments

We performed pulse experiments by pulsing 0.6 cm3 of I-butene over the powdered sample at different temperatures. Figure 5 shows the result for the catalyst BiZMo1.0206.06. The conversion of butene per pulse decreased after the first pulse. At 450°C a small increase was observed for the second pulse, but subsequently there again was a decrease in conversion. A completely different behavior was shown by the catalyst Bi2.04M00s.06 (Fig. 6). At 400°C there was no conversion during the first three pulses, but subsequently the activity increased, to reach a maximum whereafter it

I

0 eV- 525

FIG. 4. 0 1s XPS signals for Bi2M006 (b), Bi2M01.02 06.06 (a), and Bi2.WMo06.M (c) (powered sample mounted on tape).

decreased again, as observed for the first catalyst. After an intermediate reoxidation, the conversion at 450°C started again at a low level, but rose very quickly; it reached its maximum value at the fourth pulse.

During this activation process not only did the conversion increase, but the selectivity also became higher. After activation by reduction at 450°C the sample Bi2.04Mo

I

170, =v- 165 2

t 3,ev- 2

FIG. 3. Bi 4f and MO 3d XPS signals for Bi2Mo0,.

1234567

-pulse number-

FIG. 5. Conversion of I-butene during pulse experiments on Bi2MOl.mo6.w at 350 (a), 400 (b), 425 (c), and 450°C (d).


(b) Measurement of conduction after admission of pulses of I-butene (reduction) at different temperatures, together with the amount of butene converted at the same temperature (simultaneous measurement of conduction and conversion).

(c) Measurement of conduction and conversion after a reoxidation.

Temperature Dependence of the Conductivity under Air and He

For this purpose we examined two of our samples, i.e., Bi2Mol.020c.M and Biz.aMo 06.06, the active and the inactive sample, in the temperature range 350-500°C. The measurements were carried out in air. The

2 6 6 6 10 -pulse number-

measured values of conductivity can be an- alyzed using the formula

FIG. 6. Conversion of 1-butene during pulse experiments on Biz.&loOs.M at 400 and 450°C.

E u(n = a0 - exp -kT . ( 1 (la)

06.06 was measured in XPS (Tables 2 and 3). The calculated Bi/Mo ratio was now 1.10; compared to the former value of 2.27 this Plotting log u vs f gave results as plotted in represents a major change, indicating that a considerable cation migration occurred

Fig. 7, for the active and the inactive sam-

during the activation process. ple. The active sample had much lower activation energy (0.8 eV) than the inactive

The catalysts Bi2M0209 and Bi2Moj012 sample (1.97 eV). showed other characteristics. At 350 andi When we carried out the temperature de- 400°C a more or less constant level of con- pendence measurements in a helium atmo- version was observed. While at 450°C the sphere instead of air, we obtained different conversion even increased during the first results. The initial activation energy for pulses; catalyst B&M0209 reached the maximum value only after 10 pulses. It could be

both active and inactive samples were much lower than under air, viz. 0.45 and

shown that during this reduction process the catalyst decomposed; an X-ray diagram taken after the reduction indicated an ap- -5

preciable amount of Bi2MoOs to be present. A survey of the conversions is given in Ta- ‘-6 ble 1.

Electrical Conductivity Measurements

The measurements performed can be di- vided into three groups:

(a) Measurement of the temperature dependence of the conductivity in the oxidized state. From these measurements an overall activation energy for conduction

FIG. 7. Temperature dependence of the electrical conductivity of Bi2MoI,0206.~ (a) and Bi2.&So06.06 (b)

can be calculated. in air.


TABLE 4

The Activation Energy E (in eV) of the Electrical Conductivity for Samples D and E after Repeated

Heating and Cooling in a Helium Stream

Heating-cooling cycle

1 2 3 4

BizM%&m 0.45 0.46 0.67 0.81 Biz.o&O%cx 1.02 1.10 0.87 0.79

1.02 eV. After several heating-cooling periods the activation energies for active and inactive sample appeared to reach the same value (Table 4).

The time dependence of the conductivity under reducing conditions (propene 5 cm3/ min) at T = 400°C again demonstrated a marked difference between the two samples. The active sample gave an immediate rise in conductivity, while the inactive sample had a much slower increase.

After several minutes of reduction with propene the conductivity appeared to ap-

preach the conductivity of the active sample.

Electrical Conductivity after Pulsing I-Butene

Conductivity and conversion were now measured simultaneously. The time between the successive pulses of butene was fixed at 25 min because of time requirements for analyzing the reaction products.

Figure 8 gives an example (BizMoO,) of the changes in conductivity with increasing number of butene pulses for three different temperatures. In general the conductivity increased with the number of pulses passed over the catalyst. It is noteworthy, however, that between pulses a decrease of the conductivity could be observed. This decrease was typical for all measurements.

In order to avoid complicated drawings, the following figures will show only the maxima in conductivity after each pulse (compare Fig. 8b with a).

Figures 9-14 show conductivity and conversion values at 350, 400, and 450°C for,

a 1 2 3 4 5 pulse nu m bcr l

-2 .

, . ”

-3 .

-4.

-5.

I’

‘b ‘...

‘~----o----o----O-~5D

Bi2M006

b ’ &Ike FIG. 8. Conversion of I-butene (0) and electrical conductivity (A) during pulse reduction of BirMo

O6 at 350,400, and 450°C. (a) Detailed picture of the changes in conductivity. (b) Maxima of conductivity after each pulse.


1 z 3.6 *5 - pulse numDer-

FIG. 9. Conversion of 1-butene (0) and electrical conductivity (A) during pulse reduction of Bi2Mol,m 06,M at 350,400, and 450°C.

respectively, B&MO l.ldk.06, Biz.o&OOmi ,

Bi2Mo209, Bi2MosOr2, MOOS, and Bi203. Figure 9 shows that Bi2M01.&6.M already had a fairly high conductivity and a high conversion at 350°C (log u = -2.8 resp. q = 0.23 after the frrst pulse with r) the fractional conversion). During further reduction the conductivity increased somewhat, but already after a few pulses a maximum value was reached. At 450°C there was hardly any change both in conductivity and conversion. This sample Bi2Mol.0206.0s showed the highest level of conduction of all our catalysts (highest value log u = -2.4). Figure 8b shows that Bi&fOo6 had a similar behavior as the former. The conductivity values were at a slightly lower level (maximum log cr = -2.45 at 450”(J), but the conversion values were even a little higher.

Figure 10 shows some remarkable features. The Sample Bi2.&iOo6.M had a very low level of conversion at 350°C; also the conductivity was relatively low compared to the former koechlinite samples (log o = -4). At 400°C after two pulses of butene the conversion increased considerably and

at the same time the conductivity arrived at a much higher level. At 450°C the conversion reached almost the same level as for the active catalysts; again the conductivity followed this trend and became as high as the value of the conductivity for the sample Bi2Mor.~O~.~. After the runs at 45O”C, the temperature was lowered to 400°C and again butene pulses were given; there was now no initial period and from the first pulse on, the sample showed a high conversion.

The samples Bi2M0209 and Bi2M030r2 showed in principle the same behavior as the Bi&fOl.d$i.os sample although the conductivity level was lower (maximum log u = -4 for B&M0209 and log (+ = -4.5 for Bi2 M03012) (Figs. 11 and 12). Figure 13 shows the situation for MOOS which had a very low conversion and a low conductivity level (log CT = -6).

No conductivity could be measured for bismuth oxide at temperatures below

'1

-3

0 01 ,o

-4,

-5 .

-6. LI

c -o- ---&so. Lo Ad.o~h-A-450

,A/ ’ .d

,,Aoo

..o’*’ A’

A’ A’

'2

-30 I '; 2 2

350 00 A-A-A-A- ,

.20

Bi2.04Mo%D6 ,&OO

/' I ,o’ , I

0 /

/ 9' I ,

-0‘ --o-’

350 I --__ 0 ---- 0 ----*--so- 12 3 4 5

pulse number-

FIG. 10. Conversion of 1-butene (0) and electrical conductivity (A) during pulse reduction of Biz.&o O,., at 350,400, and 450°C.

VAN OEFFELEN, VAN HOOFF, AND SCHUIT

Bi2M0209

2 4 6 8 10 pulse number-

-!

.LI

. 31

.2

.ll

FIG. 11. Conversion of I-butene (0) and electrical conductivity (A) during pulse reduction and oxidation of Bi2M0209 at 350,400, and 450°C. The numbers indicate the time intervals (in minutes) between an oxygen pulse and the next butene pulse.

500°C. Reduction lowered the conductivity at 500°C (Fig. 14) and reoxidation increased it again. This is completely opposite to the behavior of the bismuth molybdates. The difference can be explained by assuming that B&03 is a p-type semiconductor while the molybdates are n-type semiconductors.

Zntermittent Reoxidation

The foregoing experiments showed that the conductivity increased by reduction while simultaneously the fractional conversion decreased. The question is now how do conductivity and conversion behave after an intermittent oxidation pulse?

The method chosen was to dose an oxygen pulse at a certain moment after passage of a butene pulse and subsequently apply a new butene pulse. The time between oxygen pulse and second butene pulse was to

!o

A

5 ^. 2 i E

0

5

FIG. 12. Conversion of 1-butene (0) and electrical conductivity (A) during pulse reduction of Bi2Mog0r2 at 350, 400, and 450°C.

2 4 6 8 10 pulse number-

3 T

2? s 5

1”

FIG. 13. Conversion of I-butene (0) and electrical conductivity (A) during pulse reduction and oxidation of MOO, at 350, 400, and 450°C. The numbers indicate the tune intervals (in minutes) between an oxygen pulse and the next butene pulse.


1 I Bi203

ii

a00

o- o c /

l-81 butfne , o;y?yn 1

1 2 3 123 -pulse number-

FIG. 14. Electrical conductivity of B&O3 at 500°C during 1-butene and oxygen pulses.

be varied. During the series of pulses there always remained a helium flow over the catalyst.

The scenario of the experiments was as follows.

(a) A series of butene pulses was fed to the reactor similarly as before (time intervals of 25 mm),

(b) an oxygen pulse was given, (c) after t = tl a new butene pulse fol-

lowed, (d) a new series of butene pulses were

added to arrive at the same situation as after the first series,

(e) again an oxygen pulse was added, and (f) after t = t2 a new butene pulse fol-

lowed; the time differences ti and t2 were varied.

Figure 15 shows the conductivity and conversion values during these experiments for the Bi2Moi.0206.M sample. After reoxidation the conductivity became much lower. Most remarkable is the fact that conversion after reoxidation was dependent on the time between the oxygen pulse and the next butene pulse, i.e., the time during which the oxygen can do its reoxidizing work.

A short time results in a high conversion, while a long time gives a lower conversion, although this is still higher than the conversion with the first butene pulse. Reoxida- tion experiments were also carried out with the BiZMo209 (Fig. 11) and Bi2M030i2 (Fig. 12) with similar results as given above.

Another type of experiment was performed with Bi2M030i2. We examined the influence of the reoxidation during longer periods under helium. At 450°C after the fourth butene pulse we kept the catalyst for 1 h in a helium atmosphere. The next butene pulse showed a high conversion, while the conductivity during the intermediate period had decreased somewhat.

With MOOS (Fig. 13) little changes were observed during reoxidation apart from a considerable decline in selectivity. This decrease in selectivity after a reoxidation pulse was also observed for the samples Bi2 Mol.ozO~.oi and Bi2Mo3012 although to a lesser degree.

Some Additional Pulse Experiments

The catalyst Bi2Moi.~Oa.~ was used to study the conductivity under reaction conditions by giving pulses of mixtures of oxygen and butene at T = 400°C.

Figure 16 gives the influence of the ratio r = (butene)/(butene + oxygen) on conductivity and conversion. It shows that the conductivity is lower the more oxygen is present in the pulse while the fractional conversion is higher for greater oxygen contents. The selectivity of the reaction increases with higher butene ratios.

The variation of the amount of butene

1

FIG. 15. Conversion of I-butene (0) and electrical conductivity (A) during pulse reduction and oxidation of Bi2Mo,,mOaM at 400°C with different time intervals between an oxygen pulse and the next butene pulse.

I -3.5.

0 E - 4

-f&5-

L I 0.25 0.50 0.75 1.0

-r~ butene/butene + O2 -+

FIG. 16. Electrical conductivity (A), conversion of 1-butene (0), and selectivity (+) during pulses of butene-oxygen mixtures with variable composition on Bi2M01.0206.06 at 400°C.

added per single pulse was studied for BiZ Mo209 at 450°C (Fig. 17), i.e., a situation in which the use of one pulse of the habitually applied size, i.e., 0.9 cm3 immediately gave a maximum conductivity that did not vary further with additional pulses. By varying the amount of butene delivered per first pulse after complete reoxidation it was tried to determine how much butene was necessary to arrive at the maximum level. Two loops were used for this purpose while moreover the butene content could be varied still further by dilution with helium. It is seen that amounts in the order of half those normally added sufficed to obtain maximal conductivity. It should be stressed that this could not be due to complete reduction of the catalyst as seen by the amount of butene converted that was still rising.

DISCUSSION

The reduction of an oxidic catalyst starts at its surface and will proceed to the inte- rior by diffusion of oxygen anions from the bulk to the exterior. As a measure of the amount reduced per unit of surface we in- -‘U troduce the variable ZQ/S where r)p is the cm3-

fractional conversion of pulse p, Zq, the FIG. 17. Relation between the electrical conductiv- sum of the conversions of p pulses and S is ity of the first pulse of butene and the volume of the the surface of the sample. Further on we . nuke (in cm9 at T = 450°C for Bi,Mo,O.. . , - _,


shall abbreviate this to Z/S, the definition of the degree of reduction.

It is helpful to estimate the maximal Z/S value for the reduction of a surface layer. Consider a layer comparable to the (Biz 02)O or the (MoOz)O layers in the bulk of y- B&M006 (koechlinite). Such a layer has about 2 x lOi oxygens/m*; if we assume that Mo4+ and zero valent Bi represent the ultimate degree of reduction for (Moo*)0 and (Bi202)0, the maximal degree of reduction shall be between 33 and lOO%, and the number of oxygens removed between 0.6 and 2 x 1019/m2. For pulses of 0.9 cm3 (containing 2.25 X lOI butene molecules), the maximal X/S should be between 0.27 and 0.9. Because in our experiments C/S never surpassed a value of 0.5, our results appear typically relevant for an incipient reduction of the catalyst’s outer layers.

The experiments shown in Figs. 8a to 13 become more relevant when electrical conductivities are plotted as a function of X/S instead of the pulse number. For all catalysts we found (T to be linearly dependent on C/S provided the degree of reduction remained below 0.1.

CJ- = a + p * as. (2)

For some catalysts, such as Moo3 and Bi2M030i2, Eq. (2) could account for all data points. However, for Bi2M0209, the electrical conductivity became constant above X/S around 0.2 although reduction


0 a02 0.06 QIO 0.14 0.18 0.22 0.26

T/S

FIG. 18. Dependency of E [e.g., Fq. (lb)] on X/S for various catalysts.

still persisted at an approximately constant level. Similar results were also obtained for the koechlinite samples (see Figs. 8-10). Equation (2) was useful as an interpolation formula for the calculation of the activation energy E at a fixed value of X/S and different temperatures. After insertion in Eq. (lb),

log CT = log u. - 5~ x 1000/T (lb)

where u is expressed in 0-r cm-‘, and E in eV we find E as a function of US.

Proceeding in this manner, we arrived at Fig. 18. The figure shows four curved lines for five catalysts, viz., Mo03, Bi2Mo3012, BizMeOg, and the overlapping koechlinite (C) and Modoped koechlinite (D). The inactive Bidoped koechlinite E could be rep- resented by a single point e(O) = 0.98 at X/S = 0 [values of e at X/S = 0 will be further indicated as s(O)].

As readily seen, the four lines followed similar patterns. The values of 40) were always highest, i.e., the conductivities were initially low. For four catalysts we found a sharp decrease in E between X/S = 0 and 0.1 after which the E - X/S lines became almost horizontal. However, the line

for BizMosOrz was horizontal over the whole range. Numerically, the average E values of the four lines were found mark- edly different; the higher the Bi/(Bi + MO) ratio, the lower E and the higher the conductivity. It should be noted that this applied already from the very beginning of the run. The position of the line appeared es- sentially defined by what happened to the catalyst in the period preceding the introduction of butene, i.e., when the catalyst was heated in He before being exposed to butene .

The horizontal parts of the lines appear to represent a steady state where oxygen migration from bulk to surface is fast enough to replenish the oxygen depletion caused by the olefin oxidation. Evidently, to prevent the formation of electrical charge differences, oxygen migration should occur simultaneously with a coun- tercurrent electron migration inward.

(Surface) 2 (Bulk). (3)

Once situated in the bulk, these electrons apparently do not contribute any more to the conductivity, an observation to which we shall return later on.

The observation, that the electrical conductivities of the various catalysts differ considerably over the entire X/S range can hardly be explained otherwise than by assuming a reversible dissociation of 02 from the catalyst. Indeed, data in Fig. 7 and Ta- ble 4 for catalysts E and D show higher values of E and therefore lower conductivities when heated in air instead of He; the conductivity decreases with the 02 pressure. It is noteworthy that there is a definite correlation between the average E values and the ratios Bi/(Bi + MO); the higher the Bi content the lower E and the higher the conductivity. We therefore postulate that O2 dissociation occurs more readily for surface oxygens bonded to Bi cations and in particular, will be more facile on (B&)0 layers. We suggest that a pair of oxygen


ions donate electrons to a group of trivalent Bi cations under formation of an oxygen molecule and some Bi atoms in a stepwise reaction such as shown below:

(Bi3+02-)20*- e (Bi1+Cl)202- + 02, (4a)

where 0 is an oxygen vacancy, followed by a rapid disproportionation

3Bi+ F1: Bi3+ + 2Bi0. (4b)

It is indeed well established [see, for in- stance, Batist et al. (8) and Grzybowska et ~1. (9)] that Bi deposits are formed if koechlinite is heated in vacuum or He at temperatures above 673 K. Evidently, a layer-like structure as koechlinite will be particularly favorably for this reaction although it might also occur on Bi2M030i2. Since Bi atoms can be supposed to evaporate from the catalyst this would also explain the striking changes in the surface Bi/Mo ratio as found by XPS for catalyst E after activation (see Table 3).

It will be helpful at this place to analyze the activity and selectivity data as obtained in this study; they are summarized in Table 1. The first point of interest is given by a comparison of conversions in continuous flow and pulse operation; for the latter activities we only consider “first pulse” conversions. It is readily seen that the activities in continuous flow are much higher than those in pulse operation. There might be two entirely different explanations for this discrepancy. One of these is directly related to what has been discussed above; since catalysts were already reduced in He before the first pulse, their activity should be lower because part of the surface sites were still inactive because of previous oxygen losses to the carrier gas. Another expla- nation can be derived from thermal effects; reactions in continuous flow are exothermic while those in pulse in the absence of oxygen are about thermoneutral. Recent studies by Burban (26) have shown that isother- micity in continuous flow cannot be established for these reactions unless the catalysts are diluted by inert diluents such

as Sic while the reactor is cooled in a fluid- ized sandbath.

Correct activities will therefore be intermediate between those found in continuous flow and in pulse operation. Another noteworthy observation is given by the dependency of the activity on the Bi/Mo ratio. Calculated per gram of catalyst, the conversion was virtually zero for XPS equal to Bi/ MO = 2.2, to become very high around 2. A further decrease in the bulk Bi/Mo ratio from 2 to 0.67 caused a proportional decrease of conversion. However, when calculated per unit surface area the conversions remained virtually constant in this range which indicates a constant number of similar sites per unit of surface in the Bi/Mo range 2 to 0.67. Indeed, XPS values for the surface Bi/Mo ratio remained approximately constant. Matsuura et al. (7) earlier reported similar conclusions. Since catalyst E with a Bi/Mo ratio >2, was inactive, the impression was gained that a Bi oxide surface is inactive towards olefins. However, considering the evidence presented in Figs. 15 and 16, this is not entirely correct; catalysts with a Bi/Mo ratio around 2 tend to become less selective under extreme cir- cumstances such as high oxygen pressures or temperatures. Figure 15 shows that this may be connected with the formation of a transient Bi-oxygen species during the reoxidation of a partly reduced Bi oxide surface layer.

Let us now try to qualitatively explain the conductivity data and more in particular the E versus Z/S relation. We shall therefore apply band theory with special empha- sis on surface band bending (see Fig. 19). As always, the electronic band shall consist of a valence band that is filled, a conduction band that is empty and a band gap between the two. Situated in the band gap are supposed to be two sets of localized energy levels, one near the top of the valence band and the other near the bottom of the conduction band. The first set, if empty, represents Mo6+ cations, the second set Bi3+ cations.


FIG. 19. Model for band structure at surface: BP+ cation energies (IX), Mo6+ cation energies (0); posi- tions are supposed to be centerlines of distribution of states.

The band model is also supposed to represent MoOJ, with only MO levels present, or B&03 where only Bi levels occur. The tran- sition from molybdenum oxide to bismuth oxide via the bismuth molybdates is assumed to be continuous by gradually re- placing one MO by two Bi. This assumption is certainly not appropriate for bulk struc- tures but might be acceptable for surfaces. The literature furnishes some quantitative information on the properties of our surface band. Grzybowska et al. (9) studied the surface band situation of MOO, by XPS. They found the top of the valence band to be at an energy around 2.8 eV. Incipient reduction revealed the presence of two levels in the band gap at 0.9 and 2.0 eV; the latter might very well be equivalent to the E = 1.8 level found by us for MoO3; see Fig. 18. According to them these UPS bands be- long to Mo4+. Batist et al. (10) used UV- Vis diffuse reflectance spectroscopy to study the surface band properties of Mo03, Bi2M020s, and BizO3. The charge transfer band for Moos was found at 2.9 eV, hence closely similar to the results of Grzybowska et al. They also noticed a shift of the edge position for Bi2M0209 to 2.58 eV. Replace- ment of MO by Bi apparently caused a nar- rowing of the band gap, the reason why Bi molybdates are bright yellow.

Let us now consider the consequences of this qualitative surface model. Any type of reduction, whether by olefin oxidation or

by oxygen dissociation, injects electrons into the surface; these electrons are placed in the localized levels of the band gap. In- creasing the reduction shall cause the Fermi level to rise in the band gap thereby diminishing e. The low-lying MO levels will be filled first and Bi levels will be occupied at higher reduction. The conductivity shall increase because the electron density of electrons in the localized levels shall become higher while also the energy difference between Fermi level and conduction band shall diminish.

Consequently, u shall rise and E shall decrease. If the localized band system is nar- row, the decrease in E might be small, but the rise in u remains observable because X/S increases; see Eq. (2). This behavior is typical for Moos and Bi2MojOr2. However, for the Bi-rich samples reduction persisted but both u and E remained constant. We postulate that this occurs when the Fermi level arrives at an energy coinciding with the lower MO levels in the bulk. Electrons can now “spill over” from surface to bulk levels. Energetically, the bulk electrons are still far below the bulk conduction band. Since, moreover, the number of levels in the bulk is vastly greater than those in the surface band, the electron density at this particular level remains virtually constant and the surface Fermi level shall become anchored at this position. The inward migration of the electrons shall give rise to an electrical double layer with the positive charge at the surface that also contains the anion vacancies. Oxygen anions shall therefore migrate outward.

Our model provides a chemical explana- tion for the kinetic “inlet’‘-“outlet” model of Keulks et al. (II), as based on experiments with oxygen isotopes. The inlet is now identified as an ensemble of Bi atoms preferably on a partly reduced (Bi202)O layer but not restricted hereto. Similar sug- gestions were published earlier by Linn and Sleight (13) and by Otsubo et al. (14). As to the outlet, because of the work of Bur- rington and Grasselli (22), Matsuura et al.


(7), and Gai (16), there seems to be general agreement about its structure; it is a surface complex containing one Bi and one MO, similar to configurations as encountered in BiZMo209. As shown earlier, our results are in reasonable agreement herewith. As far as we know, however, it has never been explained why the inlet-outlet operation is necessary at all. The common Mars-van Krevelen model (M-VK model) assumes that reduction by the hydrocarbon (outlet) occurs at the same place as the reoxidation (inlet).

We want to propose here that the inlet- outlet operation is a necessary attribute of selective oxidation because the outlet has to be a single active oxygen in order to maintain selectivity. The inlet, however, has to contain a pair of oxygen vacancies. Inlet and outlet are therefore necessarily different both as to their composition and structural type.

When summarizing our results we con- clude that most of the new information pro- duced concerns the clarification of the function of the bismuth component. We showed above that a surface (Bi202)0 plane is easily reduced in vacuum or inert gas by dissociation of molecular oxygen, a reaction that is reversible, and therefore directly related to Keulks’ inlet site. Another function is connected with the relative inactivity of bismuth oxide versus olefins; at the temperatures applied here it is almost inert in this respect. (Bi202)0 layers can therefore partly cover (Moo*)0 surfaces and form small MO-O ensembles even containing only one active oxygen. This guarantees a selective oxidation of olefins but leads to an inhibition of a local reoxidation by molecular oxygen. Note that this is equivalent to an alternative description of the outlet site. Moreover, the facile dissociation of O2 by the (Bi20z)0 layers and the high mobility of oxygen ions and electrons created thereby ensure a facile oxygen diffusion through the lattice and as a consequence the possibility

for the replenishment of oxygen-depleted surface sites.

ACKNOWLEDGMENTS

The authors thank Professor R. Metselaar for fruit- ful discussions. They also thank Dr. L. E. Firment for showing his results on the surface science of MoOl before publication. Peter Burban of the Center for Cat- alytic Science and Technology at the University of Delaware has to be especially thanked for valuable comments. This work was supported by The Nether- lands Foundation of Chemical Research (SON) with financial aid from The Netherlands Foundation for Pure and Scientific Research (ZWO). It formed the subject of part of a Ph.D. thesis by D. A. G. van Oeffe- len at the University of Technology, Eindhoven, The Netherlands (17 October 1978).

1. 2.

3.

4. 5.

6.

7.

8.

9.

IO.

II.

REFERENCES

Pluta, J., Z. Anorg. Allg. Chem. 356, 105 (1%7). Pluta, J., Z. Phys. Chem., Neue Folge 58, 189 (1%8). Peacock, J. M., Parker, A. J., Ashmore, P. G., and Hockey, J. A., J. Catal. 15, 387 (1969). Morrison, S. R., ACS preprints 22(2), 485 (1977). Berthou, H., and Jergensen, C. K., Anal. Chem. 47, 482 (1975). Batist, P. A., Bouwens, J. F. H., and Schuit, G. C. A., J. Catal. 25, 1 (1972). Matsuura, I., Schut, R., and Hirakawa, K., J. Ca- rd. 63, 152 (1980). Batist, P. A., and Lankhuyzen, S. P. P. M., J. Catal. 28, 496 (1973). Grzybowska, B., Haber, J., Marczewski, W., and Ungier, L., J. Carol. 42, 327 (1976). Batist, P. A., der Kinderen, A. H. V. M., Leeuwenburgh, Y., Metz, F. A. M. G., and Schuit, G. C. A., J. Catal. 12, 45 (1968). Keulks, G. W., J. Catal. 19, 232 (1970). Keulks, G. W., and Krenzke, L. D., in “Proceedings, 6th International Congress on Catalysis, London, 1976” (G. C. Bond, P. B. Wells, and F. C. Tom- kins, Eds.), p. 806. Chem. Sot., London.

12. Burrington, J. D., and Grasselli, R. K., J. Catal. 59, 79 (1976).

13. Linn, W. J., and Sleight, A. W., Ann. N. Y. Acad. Sci. 272, 22 (1976).

14. Otsubo, T., Miura, H., Morikawa, Y., and Shira- kasi, T., J. Cutal. 36, 240 (1975).

15. Gai, P. L., in “Fourth International Conference, The Chemistry and Uses of Molybdenum” (H. F. Barry and P. C. H. Mitchell, Eds.), p. 2%. 1982.

16. Burban, P. M., Ph.D. Thesis, University of Dela- ware, Department of Chemical Engineering, Cen- ter for Catalytic Science and Technology, 1984.

Documents

In situ measurements of the electrical conductivity of